The purpose of this project is to develop improved statistical techniques for using data from animal carcinogenicity tests conducted at high dose levels to estimate long-term risk from chemical carcinogens at very low dose levels. Existing techniques are evaluated and new techniques are studied, with particular emphasis on those which attempt to relate mechanistically to the carcinogenic process.